Google’s AI-Based Model, DeepMind’s GenCast, Dramatically Increases Weather Forecast Accuracy & Speed
GenCast is a new AI weather prediction model developed by researchers at Google’s DeepMind AI research unit. The business explained that this approach produces more accurate and timely projections up to 15 days in advance in a study that was published in Nature.
How Does It Work?
This model, which outperforms the current top system, uses probabilistic ensemble forecasting to predict a range of possible weather scenarios, giving a more comprehensive view of the conditions that will be present in the near future.
To help the larger weather forecasting community, the researchers are also making GenCast’s code, weights, and predictions publicly available. As climate change increases the frequency of severe weather occurrences, accurate forecasts are more important than ever, and this improved model may be especially important.
Unlike earlier models that provided a single “best guess,” GenCast can provide a multitude of possible weather forecasts, equivalent to 50 different weather reports. This can help you understand how the weather may change and what kinds of things might happen.
GenCast uses a special kind of AI that is very good at creating several kinds of content, including music, films, and photos. But instead of creating such things, GenCast creates different weather conditions. It’s like the AI imagining all the different ways the weather could develop.
How Is It Accurate?
Researchers fed 40 years of weather data, including temperature, wind speed, and air pressure on objects, to GenCast to teach it this. This gave GenCast a comprehensive understanding of how weather works around the world.
Researchers used meteorological data from 2019 to test GenCast’s efficacy. When they compared the model to the most popular weather forecasting system currently in use, GenCast fared better in most weather forecasts.
Researchers looked into a variety of meteorological variables, including wind and temperature, and discovered that GenCast was more accurate almost everywhere, especially when predicting the days ahead.
Advanced Features
Additionally, GenCast is adept at identifying weather uncertainty. Like erratic weather forecasts regarding precipitation, the model will reply with “maybe” instead of giving false information.
The biggest benefit is that GenCast runs really fast. In just eight minutes, it can predict the weather for fifteen days. Weather forecasting the old-fashioned way usually takes hours.